tests/testthat/test-combinatorial-phenotypes.R

In SciOmicsLab/PhenoComb: Phenotype Combinatorial Analysis

# Temporary verbose false
options(PhenoComb.verbose = FALSE)


# Read test data
# This test dataset comprises one cell for each sample for each combination
# of full-length phenotype.
# n_markers = 4
# n_samples = 4

cell_data <- read.csv("../testdata/combinatorial_phenotypes/combinatorics_test_data.csv")
channel_data <- read.csv("../testdata/combinatorial_phenotypes/channel_data.csv")
sample_data <- read.csv("../testdata/combinatorial_phenotypes/sample_data.csv")

n_markers <- nrow(channel_data)
n_samples <- nrow(sample_data)


# Generate outputs
processed_cell_data <- process_cell_data(cell_data,channel_data,sample_data)


# Test code

test_that("Generates correct number combinatorial phenotypes", {
  combinatorial_phenotypes <- combinatorial_phenotype_counts(processed_cell_data, min_count = 0)
  expect_equal(nrow(combinatorial_phenotypes), 81)
})

test_that("Generates correct counts for combinatorial phenotypes", {
  # Counts must be equal to 2^(total_markers-n_markers)
  combinatorial_phenotypes <- combinatorial_phenotype_counts(processed_cell_data, min_count = 0)
  
  row_tests <- unlist(lapply(1:nrow(combinatorial_phenotypes), function(i){
    
    considered_markers <- count_markers(as.numeric(combinatorial_phenotypes[i,1:n_markers]))
    
    return(all(combinatorial_phenotypes[i,(n_markers+1):(n_markers+n_samples)] == 2^(n_markers-considered_markers)))
    
  }))
  
  expect_true(all(row_tests))
})


test_that("Generates correct number combinatorial phenotypes using min counts", {
  filtered_combinatorial_phenotypes <- combinatorial_phenotype_counts(processed_cell_data, min_count = 5)
  expect_equal(nrow(filtered_combinatorial_phenotypes), 9)
  
  col_tests <- unlist(lapply((n_markers+1):(n_markers+n_samples), function(i){
    
    return(all(filtered_combinatorial_phenotypes[,i] >= 5))
    
  }))
  
  expect_true(all(col_tests))
  
})

test_that("Filters correctly for parent phenotypes", {
  parent_filtered_combinatorial_phenotypes <- combinatorial_phenotype_counts(processed_cell_data, min_count = 0, parent_phen = "Marker1+")
  expect_true(all(parent_filtered_combinatorial_phenotypes[,"Marker1"]==1))
  expect_equal(nrow(parent_filtered_combinatorial_phenotypes),27)
  
  parent_filtered_combinatorial_phenotypes <- combinatorial_phenotype_counts(processed_cell_data, min_count = 0, parent_phen = "Marker1+Marker2-")
  expect_true(all(parent_filtered_combinatorial_phenotypes[,"Marker1"] == 1) & all(parent_filtered_combinatorial_phenotypes[,"Marker2"] == 0))
  expect_equal(nrow(parent_filtered_combinatorial_phenotypes),9)
})


test_that("Phenotype length limit is correct", {
  for(n in 1:4){
    length_limited_combinatorial_phenotypes <- combinatorial_phenotype_counts(processed_cell_data, min_count = 0, max_phenotype_length = n)
    
    row_tests <- unlist(lapply(1:nrow(length_limited_combinatorial_phenotypes), function(i){
      
      considered_markers <- count_markers(as.numeric(length_limited_combinatorial_phenotypes[i,1:n_markers]))
      
      return(considered_markers <= n)
      
    }))
    
    expect_true(all(row_tests))
  }
  
})


test_that("Efficient mode produces same result", {
  efficient_combinatorial_phenotypes <- combinatorial_phenotype_counts(processed_cell_data, min_count = 1, efficient = TRUE)
  expect_equal(nrow(efficient_combinatorial_phenotypes), 81)
})


test_that("Multi-thread works", {
  parallel_combinatorial_phenotypes <- combinatorial_phenotype_counts(processed_cell_data, min_count = 0, n_threads = 2)
  expect_equal(nrow(parallel_combinatorial_phenotypes), 81)
  
  row_tests <- unlist(lapply(1:nrow(parallel_combinatorial_phenotypes), function(i){
    
    considered_markers <- count_markers(as.numeric(parallel_combinatorial_phenotypes[i,1:n_markers]))
    
    return(all(parallel_combinatorial_phenotypes[i,(n_markers+1):(n_markers+n_samples)] == 2^(n_markers-considered_markers)))
    
  }))
  
  expect_true(all(row_tests))
})


# Clean-up
options(PhenoComb.verbose = TRUE)

rm(cell_data,
   channel_data,
   sample_data,
   n_markers,
   n_samples,
   processed_cell_data)